| CVE |
Vendors |
Products |
Updated |
CVSS v3.1 |
| Vitess is a database clustering system for horizontal scaling of MySQL. Prior to versions 23.0.3 and 22.0.4, anyone with read/write access to the backup storage location (e.g. an S3 bucket) can manipulate backup manifest files so that arbitrary code is later executed when that backup is restored. This can be used to provide that attacker with unintended/unauthorized access to the production deployment environment — allowing them to access information available in that environment as well as run any additional arbitrary commands there. Versions 23.0.3 and 22.0.4 contain a patch. Some workarounds are available. Those who intended to use an external decompressor then can always specify that decompressor command in the `--external-decompressor` flag value for `vttablet` and `vtbackup`. That then overrides any value specified in the manifest file. Those who did not intend to use an external decompressor, nor an internal one, can specify a value such as `cat` or `tee` in the `--external-decompressor` flag value for `vttablet` and `vtbackup` to ensure that a harmless command is always used. |
| Agenta is an open-source LLMOps platform. In Agenta-API prior to version 0.48.1, a Python sandbox escape vulnerability existed in Agenta's custom code evaluator. Agenta used RestrictedPython as a sandboxing mechanism for user-supplied evaluator code, but incorrectly whitelisted the `numpy` package as safe within the sandbox. This allowed authenticated users to bypass the sandbox and achieve arbitrary code execution on the API server. The escape path was through `numpy.ma.core.inspect`, which exposes Python's introspection utilities — including `sys.modules` — thereby providing access to unfiltered system-level functionality like `os.system`. This vulnerability affects the Agenta self-hosted platform (API server), not the SDK when used as a standalone Python library. The custom code evaluator runs server-side within the API process. The issue is fixed in v0.48.1 by removing `numpy` from the sandbox allowlist. In later versions (v0.60+), the RestrictedPython sandbox was removed entirely and replaced with a different execution model. |
| The Terraform Provider for Linode versions prior to v3.9.0 logged sensitive information including some passwords, StackScript content, and object storage data in debug logs without redaction. Provider debug logging is not enabled by default. This issue is exposed when debug/provider logs are explicitly enabled (for example in local troubleshooting, CI/CD jobs, or centralized log collection). If enabled, sensitive values may be written to logs and then retained, shared, or exported beyond the original execution environment. An authenticated user with access to provider debug logs (through log aggregation systems, CI/CD pipelines, or debug output) would thus be able to extract these sensitive credentials. Versions 3.9.0 and later sanitize debug logs by logging only non-sensitive metadata such as labels, regions, and resource IDs while redacting credentials, tokens, keys, scripts, and other sensitive content. Some other mitigations and workarounds are available. Disable Terraform/provider debug logging or set it to `WARN` level or above, restrict access to existing and historical logs, purge/retention-trim logs that may contain sensitive values, and/or rotate potentially exposed secrets/credentials. |
| Spin is an open source developer tool for building and running serverless applications powered by WebAssembly. When Spin is configured to allow connections to a database or web server which could return responses of unbounded size (e.g. tables with many rows or large content bodies), Spin may in some cases attempt to buffer the entire response before delivering it to the guest, which can lead to the host process running out of memory, panicking, and crashing. In addition, a malicious guest application could incrementally insert a large number of rows or values into a database and then retrieve them all in a single query, leading to large host allocations. Spin 3.6.1, SpinKube 0.6.2, and `containerd-shim-spin` 0.22.1 have been patched to address the issue. As a workaround, configure Spin to only allow access to trusted databases and HTTP servers which limit response sizes. |
| Vikunja is an open-source self-hosted task management platform. Prior to version 2.0.0, the restoreConfig function in vikunja/pkg/modules/dump/restore.go of the go-vikunja/vikunja repository fails to sanitize file paths within the provided ZIP archive. A maliciously crafted ZIP can bypass the intended extraction directory to overwrite arbitrary files on the host system. Additionally, we’ve discovered that a malformed archive triggers a runtime panic, crashing the process immediately after the database has been wiped permanently. The application trusts the metadata in the ZIP archive. It uses the Name attribute of the zip.File struct directly in os.OpenFile calls without validation, allowing files to be written outside the intended directory. The restoration logic assumes a specific directory structure within the ZIP. When provided with a "minimalist" malicious ZIP, the application fails to validate the length of slices derived from the archive contents. Specifically, at line 154, the code attempts to access an index of len(ms)-2 on an insufficiently populated slice, triggering a panic. Version 2.0.0 fixes the issue. |
| LangChain is a framework for building LLM-powered applications. Prior to version 1.1.8, a redirect-based Server-Side Request Forgery (SSRF) bypass exists in `RecursiveUrlLoader` in `@langchain/community`. The loader validates the initial URL but allows the underlying fetch to follow redirects automatically, which permits a transition from a safe public URL to an internal or metadata endpoint without revalidation. This is a bypass of the SSRF protections introduced in 1.1.14 (CVE-2026-26019). Users should upgrade to `@langchain/community` 1.1.18, which validates every redirect hop by disabling automatic redirects and re-validating `Location` targets before following them. In this version, automatic redirects are disabled (`redirect: "manual"`), each 3xx `Location` is resolved and validated with `validateSafeUrl()` before the next request, and a maximum redirect limit prevents infinite loops. |
| LangGraph Checkpoint defines the base interface for LangGraph checkpointers. Prior to version 4.0.0, a Remote Code Execution vulnerability exists in LangGraph's caching layer when applications enable cache backends that inherit from `BaseCache` and opt nodes into caching via `CachePolicy`. Prior to `langgraph-checkpoint` 4.0.0, `BaseCache` defaults to `JsonPlusSerializer(pickle_fallback=True)`. When msgpack serialization fails, cached values can be deserialized via `pickle.loads(...)`. Caching is not enabled by default. Applications are affected only when the application explicitly enables a cache backend (for example by passing `cache=...` to `StateGraph.compile(...)` or otherwise configuring a `BaseCache` implementation), one or more nodes opt into caching via `CachePolicy`, and the attacker can write to the cache backend (for example a network-accessible Redis instance with weak/no auth, shared cache infrastructure reachable by other tenants/services, or a writable SQLite cache file). An attacker must be able to write attacker-controlled bytes into the cache backend such that the LangGraph process later reads and deserializes them. This typically requires write access to a networked cache (for example a network-accessible Redis instance with weak/no auth or shared cache infrastructure reachable by other tenants/services) or write access to local cache storage (for example a writable SQLite cache file via permissive file permissions or a shared writable volume). Because exploitation requires write access to the cache storage layer, this is a post-compromise / post-access escalation vector. LangGraph Checkpoint 4.0.0 patches the issue. |
| mchange-commons-java, a library that provides Java utilities, includes code that mirrors early implementations of JNDI functionality, including support for remote `factoryClassLocation` values, by which code can be downloaded and invoked within a running application. If an attacker can provoke an application to read a maliciously crafted `jaxax.naming.Reference` or serialized object, they can provoke the download and execution of malicious code. Implementations of this functionality within the JDK were disabled by default behind a System property that defaults to `false`, `com.sun.jndi.ldap.object.trustURLCodebase`. However, since mchange-commons-java includes an independent implementation of JNDI derefencing, libraries (such as c3p0) that resolve references via that implementation could be provoked to download and execute malicious code even after the JDK was hardened. Mirroring the JDK patch, mchange-commons-java's JNDI functionality is gated by configuration parameters that default to restrictive values starting in version 0.4.0. No known workarounds are available. Versions prior to 0.4.0 should be avoided on application CLASSPATHs. |
| A vulnerability in the Data Collection Agent (DCA) feature of Cisco Catalyst SD-WAN Manager could allow an authenticated, local attacker to gain DCA user privileges on an affected system. To exploit this vulnerability, the attacker must have valid vmanage credentials on the affected system.
This vulnerability is due to the presence of a credential file for the DCA user on an affected system. An attacker could exploit this vulnerability by accessing the filesystem as a low-privileged user and reading the file that contains the DCA password from that affected system. A successful exploit could allow the attacker to access another affected system and gain DCA user privileges.
Note: Cisco Catalyst SD-WAN Manager releases 20.18 and later are not affected by this vulnerability. |
| A vulnerability in the API user authentication of Cisco Catalyst SD-WAN Manager could allow an unauthenticated, remote attacker to gain access to an affected system as a user who has the netadmin role.
The vulnerability is due to improper authentication for requests that are sent to the API. An attacker could exploit this vulnerability by sending a crafted request to the API of an affected system. A successful exploit could allow the attacker to execute commands with the privileges of the netadmin role.
Note: Cisco Catalyst SD-WAN Manager releases 20.18 and later are not affected by this vulnerability. |
| VMware Aria Operations contains a stored cross-site scripting vulnerability. A malicious actor with privileges to create custom benchmarks may be able to inject script to perform administrative actions in VMware Aria Operations.
To remediate CVE-2026-22720, apply the patches listed in the 'Fixed Version' column of the 'Response Matrix' of VMSA-2026-0001 https://support.broadcom.com/web/ecx/support-content-notification/-/external/content/SecurityAdvisories/0/36947https:// . |
| The User Registration & Membership – Custom Registration Form, Login Form, and User Profile plugin for WordPress is vulnerable to Insecure Direct Object Reference in all versions up to, and including, 5.1.2 via the 'register_member' function, due to missing validation on the 'member_id' user controlled key. This makes it possible for unauthenticated attackers to delete arbitrary user accounts that newly registered on the site who has the 'urm_user_just_created' user meta set. |
| LORIS (Longitudinal Online Research and Imaging System) is a self-hosted web application that provides data- and project-management for neuroimaging research. Prior to versions 26.0.5, 27.0.2, and 28.0.0, an authenticated user with sufficient privileges can exploit a path traversal vulnerability to upload a malicious file to an arbitrary location on the server. Once uploaded, the file can be used to achieve remote code execution (RCE). An attacker must be authenticated and have the appropriate permissions to exploit this issue. If the server is configured as read-only, remote code execution (RCE) is not possible; however, the malicious file upload may still be achievable. This problem is fixed in LORIS v26.0.5 and above, v27.0.2 and above, and v28.0.0 and above. As a workaround, LORIS administrators can disable the media module if it is not being used. |
| Vikunja is an open-source self-hosted task management platform. Prior to version 2.0.0, a reflected HTML injection vulnerability exists in the Projects module where the `filter` URL parameter is rendered into the DOM without output encoding when the user clicks "Filter." While `<script>` and `<iframe>` are blocked, `<svg>`, `<a>`, and formatting tags (`<h1>`, `<b>`, `<u>`) render without restriction — enabling SVG-based phishing buttons, external redirect links, and content spoofing within the trusted application origin. Version 2.0.0 fixes this issue. |
| The Livemesh Addons for Beaver Builder plugin for WordPress is vulnerable to Stored Cross-Site Scripting via the `[labb_pricing_item]` shortcode's `title` and `value` attributes in all versions up to, and including, 3.9.2 due to insufficient input sanitization and output escaping. Specifically, the plugin uses `htmlspecialchars_decode()` after `wp_kses_post()`, which decodes HTML entities back into executable code after sanitization has occurred. This makes it possible for authenticated attackers, with Contributor-level access and above, to inject arbitrary web scripts in pages that will execute whenever a user accesses an injected page. |
| n8n is an open source workflow automation platform. Prior to versions 2.10.1, 2.9.3, and 1.123.22, an authenticated user with permission to create or modify workflows could leverage the Merge node's SQL query mode to execute arbitrary code and write arbitrary files on the n8n server. The issues have been fixed in n8n versions 2.10.1, 2.9.3, and 1.123.22. Users should upgrade to one of these versions or later to remediate all known vulnerabilities. If upgrading is not immediately possible, administrators should consider the following temporary mitigations. Limit workflow creation and editing permissions to fully trusted users only, and/or disable the Merge node by adding `n8n-nodes-base.merge` to the `NODES_EXCLUDE` environment variable. These workarounds do not fully remediate the risk and should only be used as short-term mitigation measures. |
| TinyWeb is a web server (HTTP, HTTPS) written in Delphi for Win32. A vulnerability in versions prior to 2.01 allows unauthenticated remote attackers to bypass the web server's CGI parameter security controls. Depending on the server configuration and the specific CGI executable in use, the impact is either source code disclosure or remote code execution (RCE). Anyone hosting CGI scripts (particularly interpreted languages like PHP) using vulnerable versions of TinyWeb is impacted. The problem has been patched in version 2.01. If upgrading is not immediately possible, ensure `STRICT_CGI_PARAMS` is enabled (it is defined by default in `define.inc`) and/or do not use CGI executables that natively accept dangerous command-line flags (such as `php-cgi.exe`). If hosting PHP, consider placing the server behind a Web Application Firewall (WAF) that explicitly blocks URL query string parameters that begin with a hyphen (`-`) or contain encoded double quotes (`%22`). |
| Due to missing authentication, a user with physical access to the device can misuse the mesh functionality for adding a new mesh device to the network
to gain access to sensitive information, including the password for admin access to the web interface and the Wi-Fi passwords.This issue affects MR9600: 1.0.4.205530; MX4200: 1.0.13.210200. |
| Due to missing neutralization of special elements, OS commands can be injected via the handshake of a TLS-SRP connection, which are ultimately run as the root user.
This issue affects MR9600: 1.0.4.205530; MX4200: 1.0.13.210200. |
| Kruise provides automated management of large-scale applications on Kubernetes. Prior to versions 1.8.3 and 1.7.5, PodProbeMarker allows defining custom probes with TCPSocket or HTTPGet handlers. The webhook validation does not restrict the Host field in these probe configurations. Since kruise-daemon runs with hostNetwork=true, it executes probes from the node network namespace. An attacker with PodProbeMarker creation permission can specify arbitrary Host values to trigger SSRF from the node, perform port scanning, and receive response feedback through NodePodProbe status messages. Versions 1.8.3 and 1.7.5 patch the issue. |
